How good are these forecasts?
Every forecast scored on seasons the model never saw while its settings were chosen.
Held-back seasons
| Forecast | Forecasts | Log loss | Favourite won |
|---|---|---|---|
| The model | 750 | 1.066 | 43% |
| Base rate | 750 | 1.087 | 41% |
| Recent form | 750 | 1.086 | 42% |
| Betting market | 749 | 1.053 | 42% |
Against the betting markets
Each market's price before the start, scored on the same finished games as the model. A negative figure means the market was more accurate.
| Market | Since | Priced | Model log loss | Market log loss | Model vs market |
|---|---|---|---|---|---|
| Bookmakers (Pinnacle, DraftKings, bet365) | 22 Jan 2026 | 430 | 1.046 | 1.026 | −1.9% |
| Polymarket | 21 Sep 2025 | 544 | 1.060 | 1.041 | −1.8% |
| Kalshi | 30 Nov 2025 | 456 | 1.052 | 1.021 | −3.0% |
Argentina plays two tournaments a year, the Apertura (January to May) and the Clausura (July to December), each a season here; thirty clubs in two zones, the top eight of each into knockout rounds.
Draws are common in Argentina, and the model's draw band is fit on the tuning tournaments.
The market comparison uses the bookmakers' closing odds with their margin taken out.
Relegation is decided on the year's combined table and the three-year points average; it is not simulated.
The knockout bracket beyond the round of 16 follows the league's usual pattern; a different draw would move the odds a little.
Questions and answers
How accurate are PodiumOracle's Liga Profesional predictions?
On 750 games from held-back seasons (Clausura 2025 to Clausura 2026), which the model never saw while it was tuned, the side it favoured won 43% of the time. Its log loss was 1.066, against 1.087 for the base rate and 1.086 for recent form, and 1.053 for the closing betting odds. Against the betting markets on the same finished games, its log loss was 1.046 against the bookmakers' 1.026 over 430 games, 1.060 against Polymarket's 1.041 over 544 games and 1.052 against Kalshi's 1.021 over 456 games (lower is better).